Kulttuurin mallintaminen: Laskennallisia tarkasteluita kulttuurisesta stratifikaatiosta

Cultural stratification refers to the social stratification of forms of cultural consumption, such as cultural taste, participation, and knowledge. Ways of consuming culture that are linked to social status maintain and reproduce social hierarchies and are thus important research subjects. In this article-based dissertation, cultural stratification, including the stratification of meanings given to the concept of culture itself, is examined computationally by rendering culture and cultural meanings measurable and modeled. The thesis draws on the theory tradition of cultural sociology influenced by French sociologist Pierre Bourdieu. Through empirical data, the focus is on the concept of culture itself as well as on cultural practices and their various forms of stratification. The dissertation focuses on modeling culture by utilizing both established and emerging computational methods in cultural sociology. In addition, it discusses the relationship between these emerging methods and quantitative methods in sociology. The question that sums up the dissertation is: What is the cultural stratification like in today's Finland and Europe, when culture is examined computationally from the perspectives of taste and cultural practices as well as the understanding of the concept of culture?

The material used in the dissertation comprise data from three surveys and one set of online discussions. Two of the survey datasets were collected in 2007 and 2018. These repeated cross-sectional surveys are representative of the Finnish mainland population (sub-articles II and III). The surveys, in addition to collecting sociodemographic factors, extensively mapped leisure activities and forms of cultural participation, taste, and knowledge (2007 n = 1388, 2018 n = 1425). The third survey (sub- article I) was conducted in 2021 with representative samples from nine European countries: Croatia, Denmark, Finland, France, the Netherlands, Serbia, Spain, Switzerland, and the United Kingdom (n = 14384). In addition to sociodemographic background factors, the survey asked about respondents’ perceptions of culture, ways of consuming culture, and, more broadly, opinions on cultural and social issues. Sub-article IV uses Finnish online discussion data. The dataset includes 2342 comments from the comment sections of Helsingin Sanomat news articles and tweets on Twitter (currently X).

The data are analyzed using three main analysis methods: structural topic modeling (sub-articles I and IV), structural equation modeling (sub-article II), and multiple correspondence analysis (sub-article III). The range of methods covers three areas of statistical research: frequentist statistics, computational analysis of text data, and geometric data analysis. In addition, principal component analysis and other dimension reduction methods, mean comparisons, and regression analyses are used as supporting methods of analysis. Thus, the dissertation uses a wide range of statistical methods and empirical data, and it includes an extensive methodological reflection. It therefore also participates in discussion on the possibilities offered by new computational methods for cultural research and sociology more broadly.

Sub-article I of the dissertation examines the meanings given to the concept of culture in different European countries and the geographical, socio-economic, and cultural-political distribution of these meanings. The meanings of culture are modeled computationally based on open answers in the survey data. The analysis shows five understandings of the concept of culture, some of which are divided into more specific topics. Based on the structural topic model of cultural meanings, the most important line separating understandings of culture is understanding it in its narrow sense as arts and, in a broad sense, as the values, customs, and traditions of different groups. In addition, culture is more commonly understood as human activity, civilization, and good manners. These understandings are primarily divided geographically but also along social and cultural-political divides. Female gender, higher age, higher level of education, higher level of cultural participation, and a left-wing political orientation toward values and economic policy are linked to a narrow understanding of culture, while connections to a broad understanding of culture are the opposite.

Part II addresses the link between social networks and cultural taste. Social networks are operationalized using the position generator method based on utilizing professional networks and prestigiousness of different professions. The position generator method provides a multidimensional perspective on measurement of social capital. Cultural taste is measured by liking different genres of music and books. According to an analysis based on structural equation modeling combining regression analysis and modeling of latent variables, broader social networks are linked to broad cultural tastes, and broader social networks are linked to a broader liking of both highbrow and popular music. The cultural capital transmitted through networks seems to contribute to the uneven distribution of cultural resources in Finland.

Part III addresses the space of lifestyles and its change in Finland from 2007 to 2018, during a period of rapid digitalization and the growing importance of social media in society. Lifestyles are operationalized as cultural practices from a total of eight fields of cultural consumption: music, TV and film, literature, food and drink, visual arts, sports and embodiment, the Internet, and everyday leisure activities. Change is examined with an identical set of variables in the first seven fields, as well as by adding variables describing Internet use and everyday leisure activities. Multiple correspondence analysis, a method from the family of geometric data analysis, is applied to study societal change. Based on the analysis, lifestyles in Finland are best structured by their legitimacy and modernity. Examined by identical cultural practices, the main structuring factors of lifestyles have remained stable, but when new fields of cultural practices are added, the main structuring dimensions of lifestyles are divided into two axes of legitimacy. The symbolic value of cultural practices is structured depending on whether they are valued from a traditional or modern perspective.

Unlike the previous sub-articles, where surveys are applied, sub-article IV utilizes online data to study discussion about Helsinki Central Library Oodi. The library expresses and distributes cultural capital and is seen as an institution undergoing transition. In many ways, Oodi represents the new roles assigned to libraries, which is why empirical material that expresses opinions on it is an interesting object of study. The textual data concerning Oodi is modeled using structural topic modeling and subsequently analyzed qualitatively using thematic close reading based on a computational text model. The results show strong polarization in attitudes toward Oodi and the supposed purpose of the library, as well as symbolic boundaries regarding the desired users and uses of the library. There is a tolerant and cosmopolitan attitude, contrasted with a more conservative view that cherishes the traditional library institution. This polarization can be seen, for example, in attitudes toward the sup- posed ownership of libraries, i.e., those seen as desired users of library services and facilities.

The dissertation makes two summarizing contributions. First, the dissertation provides new information on the central question of sociology about culture and its stratification. It shows the multidimensional social stratification of not only of cultural practices but also understandings of culture itself. The dissertation shows the stratification of culture and cultural consumption both at the individual level, such as according to individual characteristics, capitals, and values, and geographically. In addition, it takes into account temporal change in stratification. The second contribution of the dissertation is methodological. It applies methods that have so far been little or not at all used in Finnish sociology, such as multiple correspondence analysis in the study of social change or structural topic modeling in the study of the identification and distribution of socially important differences in meaning. The thesis is the first of its kind in Finnish sociology when it considers the benefits and possibilities of different empirical models of culture in the study of society and culture, especially regarding computational text analysis.